A Novel Imputation Method for Missing Values Using an RF-Based Approach
Year: 2026
Type: conference-paper
Abstract: Missing data has been a prevalent topic for decades and has posed challenges since the advent of credit cards. This paper proposes an imputation method based on Random Forest method, called XGBRF. Experiments were conducted on two real-world business datasets, where missing values generated from the merged datasets were used to train seven popular ML models. The performance was compared with three advanced imputation methods. The results indicate that the accuracy of models trained on the XGBRF imputed datasets is not only higher than that of the original datasets but also exceeds the accuracy achieved by other imputation methods.
Author: Xiaomei Feng
Institution: Macao Polytechnic University
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204946155